5.8. Avoid Common Prompt Design Mistakes

Too vague or open-ended Model doesn’t know the desired output or scope.

When a prompt is too vague, the AI model struggles to determine what exactly is expected. For example, if you ask, 'Tell me about science,' it doesn’t specify which branch of science or what particular aspect you are interested in. Without specifics, the response can be broad and unfocused, failing to address your actual needs.

Imagine asking a friend to 'tell you something interesting.' They might talk about a topic that doesn’t engage you at all. However, if you ask them to talk about interesting discoveries in astronomy, you guide them toward a response that captures your interest.

Conflicting instructions “Write a short and detailed explanation…”

When prompts contain conflicting instructions, it confuses the model. For instance, asking for a 'short and detailed explanation' sends mixed signals. A short explanation generally means being concise, while a detailed explanation requires elaboration. The model may struggle to meet both criteria, leading to unsatisfactory results.

Think about an order to a restaurant: asking for a burger that's both 'extra crispy' and 'rare' could confuse the chef. They might not know how to balance these requests without compromising one of them.

Overloading prompt Asking for too many tasks in one input.

Overloading a prompt with multiple tasks can overwhelm the model, resulting in incomplete or unsatisfactory responses. If you ask, 'Summarize this article and list three key points and provide a quote,' the model may not execute all tasks effectively because it's trying to juggle them at once.

It’s like asking someone to cook three different dishes at the same time for a dinner party. They might end up burning one dish while trying to perfect the others, resulting in an overall subpar meal.

Missing context Model lacks background to answer meaningfully.

Providing context is essential for the AI to understand and respond appropriately. If you simply say, 'Explain the process,' without specifying what process you’re referring to (e.g., manufacturing, academic research, cooking), the model lacks the necessary background, leading to vague or irrelevant answers.

Imagine a teacher saying, 'Write about the theme of the story' without mentioning which story. Students might write about different books, leading to confusion and mismatched expectations.

No formatting guidance Output may vary and be hard to parse or reuse.

Without clear formatting guidance, the AI’s output can be inconsistent and difficult to read. For instance, asking for a list but not specifying the format (numbered or bulleted) might result in a format that's hard to follow or integrate into other documents.

Consider trying to assemble furniture with vague instructions that don’t show how parts fit together. Without clear formatting, the assembly process can become confusing, resulting in a wobbly or incomplete product.

--